The relationship between motor performance and executive functioning in early childhood: A systematic review on motor demands embedded within executive function tasks

Abstract This systematic review examined to what extent response demands of executive functioning (EF) tasks influence the relationship between motor performance and EF in 2- to 6-year-old typically developing (TD) children and children with motor coordination difficulties (MCD). Eighteen of the included articles focused on TD children only and three also on children with MCD. EF tasks were subdivided based on the type of responses (i.e., motor or verbal). EF tasks requiring a motor response were subdivided into two levels (i.e., complex or simple). Results showed that the relationship between motor performance and EF in 2- to 6-year-old TD children and children with MCD was inconclusive with the strength of correlation coefficients for the most part varying from very weak to moderate. The type of EF task response did not influence the relationship between motor performance and EF. The review thus implies that other task demands than the type of EF task response need to be investigated to explain the inconsistent relationship between motor performance and EF, such as the complexity of the motor response of EF tasks, the complexity of the actual motor tasks, and additional processes (e.g., memory, intelligence, language comprehension).


Introduction
Early childhood, here defined as the period from birth to 6 years of age, is an important and fundamental period of rapid development in human life (McCartney & Phillips, 2006).Two especially important developmental domains that show substantial development during early childhood are motor performance and executive functioning (EF; Diamond, 2006;Gabbard, 2018).As domains, they are vital to cognitive and social development (Leonard & Hill, 2014;Moriguchi, 2014;Veldman et al., 2019), and predict school readiness (Blair & Razza, 2007;Grissmer et al., 2010).Motor performance can be described as "the observable production of a voluntary action or a motor skill" (Schmidt & Wrisberg, 2008, p. 11).It includes planning, organizing, monitoring, and controlling motor coordination (Roebers & Kauer, 2009).EF is defined as the ability to plan, coordinate, regulate, and monitor action control and thought using higher-order cognitive processes (Carlson et al., 2016).A relationship between motor performance and EF seems selfevident from a conceptual point of view as they both concern action control, such as monitoring, planning, and sequencing of actions (Koziol et al., 2012).The relationship can also be understood through embodied cognition theories, which suggest that cognitive development, including EF, occurs in the context of the individual's bodily interactions with the environment (Barsalou, 2008;Marmeleira & Duarte Santos, 2019).
Interestingly, empirical studies in typically developing (TD) children have shown inconclusive results regarding the relationship between motor performance and EF.A recent meta-analysis focused on the relationship between motor performance and EF in 3-to 12-year-old TD children (Gandotra et al., 2022).Small effect sizes were found on the one hand for a global relationship, and on the other hand for relationships between specific components of motor performance (i.e., manual dexterity, locomotor skills, balance, and object control) and EF (i.e., inhibition, working memory, and cognitive flexibility).Manual dexterity and balance were significantly weakly related to all EF components (i.e., inhibition, working memory, and cognitive flexibility), whereas locomotor skills were only significantly weakly related to working memory.Non-significant relationships were found between object control skills and the EF components (Gandotra et al., 2022).Of note is that the included individual studies showed highly varying correlation coefficients, ranging from zero to moderate.Gandotra et al. (2022) also examined whether age moderated the relationship between motor performance and EF, and thereby, whether age could explain the highly varying correlations.Results of the meta-regression showed that age affected only the component-specific relationship between balance and inhibition.More specifically, the relationship between balance and inhibition became stronger with age.No other moderating effects were examined.It was, however, evident that the reviewed studies applied vastly different motor and EF measures.Hence, the high heterogeneity of motor and EF measures used in these studies may explain the highly varying correlations.
There are several studies that take a clinical perspective into examining the relationship between motor performance and EF (Leonard & Hill, 2014;Bernardi, 2018).For example, researchers have examined conditions in which motor performance is disrupted, like in the case of Developmental Coordination Disorder (DCD).It provides an ideal opportunity to investigate relationships between motor performance and EF as it is a disorder diagnosed based on difficulties in acquiring and executing motor skills (APA, 2013).However, it is important to note that, although motor coordination difficulties are the defining features of children with DCD, heterogeneity exists in the nature and severity of the motor coordination difficulties (Asonitou & Koutsouki, 2016;Green et al., 2008;Lust et al., 2022).For example, subtypes of DCD can be distinguished based on more severe difficulties in fine motor performance, gross motor performance, or both (Lust et al., 2022).Research has suggested that many children with DCD have difficulties not only in motor performance but also in EF (Lachambre et al., 2021), yet these difficulties do not always co-occur (Goulardins et al., 2016;Molitor et al., 2015).In addition, a recent systematic review examining the relationship between motor performance and EF in individuals aged 5 years and older with DCD showed highly varying correlation coefficients, ranging from weak to strong (Fogel et al., 2021).Put concisely, an indecipherable picture emerges from the empirical studies available in both TD children and children with DCD.Therefore, more studies are needed to disentangle possible reasons for the unclarity.One of the recommendations for further research by Fogel et al. (2021) was to pay attention to the different types of assessment used to examine the relationship between motor performance and EF.Our review responds to this call by carrying out a detailed methodological inspection, with the hypothesis that shared motor task demands have an effect on the association between motor performance and EF.
One of the most frequently cited challenges of assessing EF is that multiple processes, both EF and non-EF-related (such as verbal or motor responses), contribute to interindividual differences in the actual EF task performance (Zelazo et al., 2016).Consequently, children's performance on an EF task does not result solely from their EF skills, but also from other processes that are necessary to perform the task.This phenomenon is commonly known as the "task impurity problem" (Friedman & Miyake, 2017).In early childhood, EF domains come to the fore in a conjoint way as they are still in the process of developmental differentiation (Shing et al., 2010).Concurrently, other non-executive processes that are necessary to perform EF tasks are still in development (Wasserman & Wasserman, 2013).For clinicians and researchers, it is thus difficult to discern whether a low score on an EF task is due to poor EF, or due to some other child-or task-related difficulties (e.g., motor performance, processing speed, visual perception; Aadland et al., 2017;Michel et al., 2018).For these reasons, the potential confounding influence of other (non-executive) processes on EF measurement is of particular concern in early childhood (Wasserman & Wasserman, 2013).
Notably, many EF tasks require to some degree aspects of motor performance (McClelland & Cameron, 2019).For example, fine motor skills are required in tasks such as the TRAILS-P in which children are asked to draw a trail from the smallest mouse to the biggest mouse, connecting with each mouse's cheese before proceeding to the next mouse (Monette et al., 2015), or the Knock-tap task in which children are asked to respond by knocking when the tester taps the table and vice-versa (Luria, 1980).Obviously, EF tasks with motor demands are useful for assessing the ability to deploy EF in the motor domain, but they might be less useful for investigating whether motor performance and EF are related.That is, the relationship between motor performance and EF tasks requiring a motor response might be attributed to their overlap in motor demands, rather than the pure involvement of EF in motor performance (see Kaushanskaya et al. [2017] for a similar argument regarding verbal demands).If so, poor performance on EF tasks with motor demands could actually arise from underlying motor coordination difficulties.However, as this may seem obvious, results in children with DCD have not been clear-cut.For example, Pratt et al. (2014) found that children with DCD performed significantly worse than TD children on the inhibition task with a verbal response, but not on the inhibition task with a motor response.In addition, no significant correlations between motor performance and any of the EF tasks were found in either the children with DCD or TD children.In contrast, Leonard et al. (2015) showed that children with DCD and children who demonstrated motor coordination difficulties during the screening of a normative sample performed significantly worse on working memory, inhibition, planning, and fluency, but only in tasks requiring a motor response.The mixed results suggest that the relationship between motor performance and EF is not explained by the type of EF task response (i.e., motor or verbal) only.
Among possible mechanisms influencing the relationship between motor performance and EF, the level of automaticity of motor demands in EF tasks may also play a role.The theory of automaticity suggests that tasks involving automatized and well-learned motor skills require limited EF, but that EF is especially required in motor tasks that are either novel, require a fast response, or when conditions and demands of a motor task change (Ackerman, 1988;Diamond, 2000;Floyer-Lea & Matthews, 2004).In a similar vein, EF tasks requiring a complex motor response, in other words with a higher cognitive load, may thus be more difficult than EF tasks requiring a simple motor response.Consequently, motor performance may be more stronger related to EF tasks requiring complex motor responses than to EF tasks requiring a simple motor response due to a higher overlap in cognitive load between the motor task and the EF task requiring a complex motor response.Therefore, the level of automaticity of motor demands in EF tasks should be taken into account when examining the unique role of the EF task response demands on the relationship between motor performance and EF.
The indication of the level of automaticity of motor demands in EF tasks is challenging because there are no clear guidelines to indicate the level of automaticity of motor demands.The conceptualization of task difficulty by Guadagnoli and Lee (2004) can be used for this.Guadagnoli and Lee (2004) differentiated between nominal and functional task difficulty.Nominal task difficulty refers to the characteristics of a task independent of the person performing it.Functional task difficulty refers to the level of difficulty taking the skill level of the person performing the task into consideration (Guadagnoli & Lee, 2004).The level of an EF task's motor response is, thus, dependent on the characteristics of the task and on the skill level of the child.The skill level of a person is also dependent on age.In early childhood, concomitant to the ongoing development, some motor demands are experienced as more difficult than other motor demands.For example, pencil skills are not yet fully developed in early childhood (Lin et al., 2017), which may hamper the actual execution of the EF task.However, motor demands, such as pointing or pushing a button are automated in most young children and may only have a limited or neglectable influence on the execution of EF tasks.That said, large variability in motor performance may exist between young children, complicating the impact of functional task difficulty even further.
In sum, there is previous literature examining the relationship between motor and EF in TD children and children with DCD.The recent meta-analysis of Gandotra et al. (2022) examined the relationship between motor performance and EF in 4-to 16-year-old TD children.In addition, the systematic review of Fogel et al. (2021) examined their relationship in individuals from the age of 5 years with DCD.However, the role of task demands that could explain the highly varying results between the empirical studies was not the focus of these papers.This systematic review, therefore, examined the role of EF task response demands on the relationship between motor performance and EF in early childhood.The focus of the current review lies on early childhood because this time period is increasingly recognized as a foundational period for developing motor, cognitive, social-emotional, and academic skills (Gabbard, 2018;Gilmore et al., 2018;McCartney & Phillips, 2006).Notwithstanding the suggestion that EF emerges in infancy (Hughes, 2011), the focus of this review is on motor performance and EF of 2-to 6-year-old children.The age range is thus beyond this early period because infant measures of EF are often not specific enough to distinguish between the specialized EF skills considered for this review (Mulder et al., 2009).To examine whether there exists a relationship between motor performance and EF or whether their relationship could be attributed to the overlap in motor demands, it is essential to include the full spectrum of motor performance, that is, children with age-typical motor skills as well as children with motor coordination difficulties (MCD).Therefore, the current review focusses on both TD children and children with MCD.We use the term MCD in this review, as current international DCD guidelines do not recommend giving a diagnosis before the age of 5 (Blank et al., 2019), and a further limitation is that empirical studies focusing on young children do not regularly assess all DSM-5 criteria for DCD (Li et al., 2021).Based on the diagnostic guidelines for DCD (Blank et al., 2019), we considered children as having MCD when (1) they scored at or below the 15th percentile on a standardized motor test, 1 and (2) they did not have any neurological and/or medical conditions, sensory impairments, intellectual disabilities, and/or were born preterm.The results of this review can provide clinicians and researchers with a critical perspective on the choice of EF measures in the context of clinical and research diagnostics and intervention.In addition, the results can provide information that is relevant for early education when developing and evaluating targeted interventions related to motor performance and EF.More specifically, it provides information on whether early education programs and interventions should focus on only one or both developmental domains (i.e., motor and EF domains).It was hypothesized that due to the involvement of motor performance in EF tasks requiring a motor response, performing EF tasks that require such a response is more strongly associated with motor performance than when executing EF tasks requiring a verbal response.

Literature search and selection criteria
The literature search was conducted on August 13, 2021, using PsycInfo, PubMed, and Web of Science.The search strategy differed per database.We developed a search strategy for PsycInfo using Thesaurus [DE], title [TI], and abstract [AB] terms, searching for the terms (1) motor performance in combination with (2) executive functions.In addition to the general term executive functions, inhibition, working memory, and cognitive flexibility were used as search terms, because it is broadly agreed that these are the core components of EF (Diamond, 2013;Miyake et al., 2000).Publication dates between January 2000 and August 2021, and age group 0-12 years were used as the initial search focus.For PubMed, a search strategy was developed using MeSH (medical subject heading) and tiab (title/ abstract) terms, searching for the terms (1) motor performance in combination with (2) executive functions, and (3) children.Publication date between 2000 and 2021 was used as a search focus.In Web of Science, papers were searched using Topic terms, searching for the terms (1) motor performance in combination with (2) executive functions, and (3) children.Full details of the searches can be found in Supplementary Appendix A.
Inclusion criteria for this review were that the studies: (1) were published between 2000 and 2021, (2) were written in English or German, (3) focused on TD children and/or children with MCD aged 2-6 years old, (4) measured motor performance and EF using performance-based measures, (5) reported on the relation between motor performance and EF using analyses that test associations, i.e., a type of (canonical) correlation, regression, or factor structure, and (6) reported original empirical data.The onset for the systematic search, i.e. the year 2000, was deliberately chosen because the interest in the examination of the relationship between motor performance and EF has expanded since the article of Diamond was published (2000).Studies focusing on children that scored at or below the 15 th percentile on a standardized assessment of motor performance were included as MCD samples, including a version of the Movement Assessment Battery for Children (MABC; Henderson et al., 2007;Henderson & Sugden, 1992), McCarron Assessment of Neuromuscular Development (MAND;McCarron, 1997) or Bruininks Test of Motor Proficiency-2 (BOT-2; Bruininks & Bruininks, 2005).
Exclusion criteria for this review were: (1) reviews, metaanalyses, and conference abstracts, (2) studies with participant groups having neurological and/or medical conditions, sensory impairments, intellectual disabilities, and/or born preterm, (3) studies only presenting longitudinal data, (4) intervention studies, (5) studies using only observation or rating measures, and (6) studies that included confounders, covariates, mediators, or moderators in the analyses when examining the relationship between motor performance and EF without presenting raw-order analyses.In case the raw-order or cross-sectional association data between motor performance and EF were not reported, the authors were contacted to provide these.
After the removal of duplicates, title screening of 10% of randomly selected titles was performed by two reviewers (GV and SH) to calculate interrater agreement with Cohen's kappa.A Cohen's kappa score of .70 was set as the accepted standard (Penman, 1980).Disagreements between both reviewers about including papers based on titles were resolved by discussion.A third reviewer (MH) was consulted when a consensus could not be reached.Because the agreement on this title screening was almost perfect (97.8% agreement; j ¼ .858,p < .001), the remaining titles were divided and screened by the two reviewers separately.Next, the abstract screening of 10% of randomly selected abstracts was performed by two reviewers (GV and SH).The interrater agreement of the abstract screening was substantial (94.19% agreement; j ¼ .821,p ¼ .001).Therefore, the abstracts of the selected papers based on the titles were divided and reviewed by the two reviewers separately.Next, full-text screening of 10% of randomly selected full-texts was performed by two reviewers (GV and SH).The interrater agreement of the full-text screening was perfect (100% agreement; j ¼ 1.000, p < .001).Thus, the full texts of the selected papers based on the abstracts were divided and assessed for eligibility by the two reviewers separately.

Methodological quality
The studies included in the systematic review were assessed with regard to methodological quality using the Critical Review Form-Quantitative Studies (Law et al., 1998).Two questions of the Critical Review Form-Quantitative Studies were excluded from this methodological quality analysis.The question "Were there any biases?"was excluded because all studies include potential biases (Kazdin, 2016).The question "was the intervention described in detail?" was excluded because intervention studies were excluded from this review.For the same reason, these questions have been excluded in other reviews as well (e.g., Lanctôt & Guay, 2014).The questions were scored as "yes" (1 point), "no" or "not reported" (0 points).Question number 12 (Were drop-outs or withdrawals reported?)was scored reversed.Scores for all 15 questions were summed for each study.As a rule of thumb, a total score of or below seven points was considered as a poor methodological quality, a total score between eight and eleven points was regarded as having average methodological quality, and a total score of twelve points or higher was considered as high methodological quality.All appraisals of retrieved studies were primarily undertaken by the first author (GV) and independently verified by the last author (SH) for 25% of the studies.Disagreements were resolved by discussion and, if needed, with the assistance of the second author (MH).The agreement on the appraisals was almost perfect (92.9% agreement; j ¼ .882,p < .001).

Data extraction and analysis
The following data were extracted from the included studies: sample size, age, gender, developmental status (i.e., TD or MCD), motor and EF measures, motor and EF component(s) measured, type of EF task response (motor or verbal), and, when applicable, the complexity of the motor response of an EF task (complex or simple).
Motor performance was divided into the following six components: fine manipulative skills (e.g., cutting, drawing), gross manipulative skills (e.g., throwing, kicking), locomotor skills (e.g., running, jumping), stability (e.g., standing on one leg), body coordination (e.g., jumping laterally), and total motor performance (i.e., the score of a combination of motor components; Gabbard, 2018).These components have been used before in a review by Barnett et al. (2016).Since aspects of motor performance are not always fully distinguishable from each other, some motor skills could be classified under multiple components.In such cases, they were classified based on the most prominent type of motor category required to carry out the task.
EF tasks were divided based on its core components (Diamond, 2013;Miyake et al., 2000) into the following components: inhibition, working memory, and cognitive flexibility.In addition, a fourth category, "other" (such as planning) was used.Depending on the type of response of the EF task, they were further subdivided into EF tasks requiring a verbal response, a complex motor response, or a simple motor response.Following the theory of automaticity (Floyer-Lea & Matthews, 2004) and the conceptualization of task difficulty (Guadagnoli & Lee, 2004), motor responses of EF tasks were differentiated into complex and simple motor responses.Motor responses to EF tasks are regarded as "complex" when they are less practiced and less automated than simple motor responses (Willingham, 1998).Motor responses, such as drawing a line and writing were categorized as complex motor responses of EF tasks in young children, and motor responses, such as pressing a button, pointing, and touching blocks were categorized as simple motor responses of EF tasks in young children.Due to different types of scoring of motor performance and EF tasks, a negative correlation coefficient can reflect a positive or a negative relationship.For example, when motor performance is scored by accuracy and EF is scored by the response time, then a negative correlation coefficient reflects a positive relationship between motor performance and EF (i.e., better motor performance is related to better EF).To keep the interpretation of all correlations in the same direction, namely positive correlations reflecting positive relationships, we reversed the sign of a negative correlation coefficient to a positive one in case a negative correlation coefficient reflected a positive relationship and vice versa.The absolute values of correlation coefficients above .50were considered as strong, correlations between .30and .50 were considered moderate, correlations between .10 and .30were considered weak, and correlations equal to or below .10 were considered as no correlation (Cohen, 1988).P-values were ignored because they are difficult to interpret due to sample(-size) dependency, design power issues, and the (variation in the) reliability of the measures being used (Cohen, 1990;Cumming, 2014).
To draw reliable and valid conclusions from the reviewed articles with regard to the effect of the type of EF task response on the strength of relationships between motor performance and EF, we used a minimum of five correlations per area (i.e., motor and EF component, type of EF task response, or level of motor response) as a rule of thumb.Differences between the strength of correlations examined with one type of EF task response and the strength of correlations examined with EF tasks requiring another type of EF task response were inspected by evaluating every strength of the correlation category separately.To provide a detailed explanation of the decision on when we determined a difference between both types of task responses, three examples are shown in Supplementary Appendix B.
All data extraction was undertaken by the first author (GV) and independently verified by the last author (SH) for 25% of the studies.The agreement on the data extraction was almost perfect (98.1% agreement; j ¼ .981,p < .001).
In the Results section, we use the terms "motor scores" and "EF tasks."The term motor scores refer to task-, component-and/or total scores.More specifically, task scores refer to scores gained from a specific task.Component scores refer to scores gained from multiple tasks that are aggregated into one score reflecting a component of motor performance as defined in this review (i.e., fine manipulative skills, gross manipulative skills, locomotor skills, stability, or body coordination).Total scores refer to scores gained from multiple tasks aggregated into one score reflecting multiple motor components.The term EF tasks refer to only task scores.

Search
The search identified a total of 3,981 records (see Figure 1).After removing duplicates, a total of 2,690 records were identified.An initial screening of titles reduced the number of articles to 172.After completing the abstract review, 67 potentially relevant articles were identified.Based on full text and in-and exclusion criteria, 34 articles were selected for this systematic review.Correlations (i.e., zero-order, cross-sectional, or per task) were requested from authors for 12 articles.Eight of these 12 articles were excluded because correlations were not provided after request.Five articles were excluded because they turned out to be additional duplicates already included in the review.This resulted in a total of 21 articles.Table 1 shows the study characteristics of the included articles.Out of these 21, 18 studies focused on TD children only, and three studies focused on both children with MCD and TD children.In this review, the results are described per developmental status.

Methodological quality
Eighteen (85.7%) of the 21 studies showed average methodological quality (see Table 2).One study (4.8%) showed poor methodological quality and two studies (9.5%) showed high methodological quality.All studies scored well on the quality standards regarding study purpose, review of literature background, and evidence testing.Most studies scored well on quality standards regarding design (95.2%), analysis methods (66.7%), conclusion (85.7%), and limitations (90.5%).Methodological limitations identified in many studies were the lack of detailed description of the sample (57.1% of studies did not clearly report this information), reporting of clinical importance (57.1%), and reporting of clinical implications (57.1%).The reliability of the used measures was mostly not reported (71.4%).Only one study reported the validity of measures.None of the studies justified the sample size.

Study characteristics
Table 3 shows that seven studies used a version of the Movement Assessment Battery for Children (MABC).The other studies used a large variety of motor measures.Some studies used one test battery, a combination of test batteries, a single motor task, or multiple separate motor tasks.As can be seen in Table 4, there was also a large variation in EF measures, both between-and within-studies.Most studies focused on multiple EF components.
The type of response of EF tasks and the relationship between motor performance and EF.Of the reported correlations, 74.5% were correlations between motor scores and EF tasks requiring a motor response.Of note, motor scores refer to task-, component-, and/or total scores; EF tasks refer to only task scores.Of the correlations between motor scores and EF tasks requiring a motor response, 3.3% Table 1.Study characteristics of the included articles.Children who scored at or below the 15 th percentile on the total score of the Movement Assessment Battery for Children Second Edition (MABC-2; Henderson et al., 2007) were identified as at risk for DCD.

References
b Children who scored below the tenth percentile on the manual dexterity subtest of the MABC-2 were included in the motor coordination impairment group. c Children who scored below the tenth percentile on the total score of the MABC-2 were included in the motor coordination impairment group.
Table 2. Methodological quality of the reviewed studies."yes" ¼ 0 points, "no" ¼ 1 point.strongly positive, 23.8% moderately positive, 49.7% weakly positive, 21% no, and 2.2% weakly negative correlations were reported.Of the correlations between motor scores and EF tasks requiring a verbal response, 8.1% strongly positive, 24.2% moderately positive, 37.1% weakly positive, 25.8% no, and 4.8% weakly negative correlations were reported.As can be seen in Figure 2, looking at all motor and EF scores, both types of EF responses (motor and verbal) resulted in mostly weakly positive correlations between motor and EF scores (49.7 and 37.1%, respectively).Furthermore, we compared the correlations between specific motor and EF components.As can be seen in Figure 3, the correlation between fine manipulative skills and inhibition was higher when EF tasks requiring a motor response were used.The relationship between fine manipulative skills and working memory did not show differences in the strength between EF tasks requiring a motor response and EF tasks requiring a verbal response (see Figure 3).Regarding the relationship between gross manipulative skills and inhibition, the results are inconclusive when comparing the strength of correlations examined with EF tasks requiring a motor response with the strength of correlations examined with EF tasks requiring a verbal response (see Figure 4).As can be seen in Figure 4, the correlation between gross manipulative skills and working memory was higher when EF tasks requiring a verbal response were used.The relationships between stability on the one hand and inhibition and working memory, on the other hand, were higher when EF tasks requiring a motor response were used than when EF tasks requiring a verbal response were used (see Figure 5).

Study characteristics
Table 3 shows that all three studies used a version of the MABC.As can be seen in Table 4, there was a large between-and within-studies variation in EF measures.All three studies focused on inhibition and working memory.Two studies also focused on cognitive flexibility and one study on verbal fluency.

Outcomes
The studies on children with MCD reported 26 correlations between motor scores and EF tasks.The relationship between motor and EF scores varied largely with correlations ranging from À.362 to .619.Mostly moderately positive (26.9%), weakly positive (26.9%), and no correlations (30.8%) were reported.
The type of response of EF tasks and the relationship between motor performance and EF.Of the reported correlations, 76.9% were correlations between motor scores and EF tasks requiring a motor response.As can be seen in Figure 6, looking at all motor and EF scores, both types of  EF responses (motor and verbal) resulted in highly varying correlations.The effect of the type of EF response on the relationship between specific motor and EF components could not be examined, because there were few or no correlations between specific motor and EF components per type of EF response.

Main findings
This systematic review examined the role of EF task response demands on the relationship between motor performance and EF in 2-to 6-year-old TD children and children with MCD.Our review showed that the reported correlations varied noticeably in terms of strength, with most of the correlations (92.6% in typically children, 84.6% in children with MCD) ranging from none to moderately positive.Also, negative correlations were found (8.3% in TD children and 15.4% in children with MCD), indicating that better motor performance is related to worse EF.In contrast to our hypothesis, the strength of the relationship between motor performance and EF did not seem to differ between the use of EF tasks requiring a motor response and EF tasks requiring a verbal response.This was found in both TD children and children with MCD.It must be highlighted that for both types of EF task responses (i.e., motor or verbal), highly varying relationships with motor performance were found.In sum, previous research has found that task demands may strengthen or attenuate the relationship between motor performance and EF (Houwen et al., 2017;Roebers & Kauer, 2009).However, as shown in our systematic review, the full explanation is challenging to tease out.
One of the reasons why the type of response demands used in specific EF tasks (i.e., motor or verbal) did not explain the varying relationships between motor performance and EF could be that motor demands included in EF tasks requiring a motor response are less apparent than expected.As it happens, only one of the included studies in this review used an EF task requiring a complex motor response, namely the NEPSY visual attention subtest in which children were asked to circle certain pictures using a pencil or crayon (Kim et al., 2018).All other studies included EF tasks requiring a simple motor response, such as pointing at cards, touching blocks, or pressing a button.These simple movements start developing already in infancy (Camaioni et al., 2004) and can be assumed to be well-practiced and automated in (TD) preschool children.Following the theory of automaticity (Floyer-Lea & Matthews, 2004), the simple motor response of the EF tasks allows children to engage in the actual motor execution with a minimal cognitive load and, therefore, no clear differences in this review may have been found between EF tasks with motor demands vs. EF tasks with verbal demands.Due to only one correlation with an EF task requiring a complex motor response in the data, the current review could not examine the differential effect of the level (i.e., complex or simple) of the motor response of EF tasks on the relationship between motor performance and EF.
It is important to note that in TD children and children with MCD, the relationship between motor performance and EF has been examined clearly more often using EF tasks requiring a motor response than using EF tasks requiring a verbal response.As a consequence, it is complicated to disentangle the true effects of the type of EF task response on the relationship between motor performance and EF.Correlations between all motor and EF scores per type of EF task response in typically developing children.Note.Positive correlations reflect a positive relationship, that is, better motor performance is related to better EF.Negative correlations reflect a negative relationship, that is, better motor performance is related to worse EF.The absolute values of correlation coefficients above .50or below À.50 were considered as strongly positive or negative, respectively, correlations between .30and .50 and between À.30 and À.50 were considered moderately positive or negative, respectively, correlations between .10 to .30and À.10 to À.30 were considered weakly positive or negative, respectively, correlations equal to or below .10 and equal to or below À.10 were considered as no correlation (Cohen, 1988).Of the EF tasks requiring a motor response, 2.2% showed a weakly negative correlation, 21.0% showed no correlation, 49.7% showed a weakly positive correlation, 23.8% showed a moderately positive correlation, 3.3% showed a strongly positive correlation with motor scores.Of the EF tasks requiring a verbal response, 4.8% showed a weakly negative correlation, 25.8% showed no correlation, 37.1% showed a weakly positive correlation, 24.2% showed a moderately positive correlation, and 8.1% showed a strongly positive correlation with motor scores.
Nevertheless, it is surprising that studies in children with MCD mainly used EF tasks requiring a motor response, although these children have an extra difficulty in performing EF tasks with motor demands due to their motor coordination difficulties (Leonard et al., 2015).To add to the complex picture of findings, children with MCD are more impaired in EF tasks requiring a motor response than in those requiring a verbal response (Lachambre et al., 2021).
Besides the response demand of EF tasks, it could also be argued that the relationship between motor performance and EF may depend on the complexity of the actual motor task.For example, Maurer and Roebers (2019) showed stronger relationships between complex motor tasks (i.e., adapted tasks of the MABC-2 and KTK) and EF than between simple motor tasks (i.e., original tasks of the MABC-2 and KTK) and EF in children aged 5-6 years.Positive correlations reflect a positive relationship, that is, better motor performance is related to better EF.Negative correlations reflect a negative relationship, that is, better motor performance is related to worse EF.The absolute values of correlation coefficients above .50or below À.50 were considered as strongly positive or negative, respectively, correlations between .30and .50 and between À.30 and À.50 were considered moderately positive or negative, respectively, correlations between .10 to .30and À.10 to À.30 were considered weakly positive or negative, respectively, correlations equal to or below .10 and equal to or below À.10 were considered as no correlation (Cohen, 1988).Positive correlations reflect a positive relationship, that is, better motor performance is related to better EF.Negative correlations reflect a negative relationship, that is, better motor performance is related to worse EF.The absolute values of correlation coefficients above .50or below À.50 were considered as strongly positive or negative, respectively, correlations between .30and .50 and between À.30 and À.50 were considered moderately positive or negative, respectively, correlations between .10 to .30and À.10 to À.30 were considered weakly positive or negative, respectively, correlations equal to or below .10 and equal to or below À.10 were considered as no correlation (Cohen, 1988).
Although empirical correlational studies are interesting and suggest that motor performance and EF are related, such task-specific approaches also pose problems.That is, neither motor performance nor EF can be assessed in isolation.In addition to targeting a particular executive function, EF tasks may also require a motor or verbal response as well as other unintendedly measured processes, such as memory and language comprehension (Jurado & Rosselli, 2007).Measuring EF is also further challenged by the fact that EF tasks usually tap multiple EF components simultaneously (Friedman & Miyake, 2017).For example, it has been suggested that the Head-Toes-Knees-Shoulders task is an EF task including three types of EF, i.e., inhibition, working memory, and cognitive flexibility (McClelland et al., 2014).In a similar vein, it can also be argued that different motor tasks entail distinct additional processes, such as attention and memory (Song, 2019), which could make the motor tasks more difficult for instance for children with attention difficulties.Understandably, this highly challenges the interpretation of (poor) performance on motor and EF tasks.Other additional processes that are being measured by motor and EF measures may play a prominent confounding Positive correlations reflect a positive relationship, that is, better motor performance is related to better EF.Negative correlations reflect a negative relationship, that is, better motor performance is related to worse EF.The absolute values of correlation coefficients above .50or below À.50 were considered as strongly positive or negative, respectively, correlations between .30and .50 and between À.30 and À.50 were considered moderately positive or negative, respectively, correlations between .10 to .30and À.10 to À.30 were considered weakly positive or negative, respectively, correlations equal to or below .10 and equal to or below À.10 were considered as no correlation (Cohen, 1988).Figure 6.Correlations between all motor and EF scores per type of EF task response in children with MCD.Note.Positive correlations reflect a positive relationship, that is, better motor performance is related to better EF.Negative correlations reflect a negative relationship, that is, better motor performance is related to worse EF.The absolute values of correlation coefficients above .50or below À.50 were considered as strongly positive or negative, respectively, correlations between .30and .50 and between À.30 and À.50 were considered moderately positive or negative, respectively, correlations between .10 to .30and À.10 to À.30 were considered weakly positive or negative, respectively, correlations equal to or below .10 and equal to or below À.10 were considered as no correlation (Cohen, 1988).
role in the relationship between motor performance and EF.It thus is important to take other additional processes into account when examining the relationship between motor performance and EF.Indeed, several studies have attempted to examine the influence of confounding additional processes on the relationship between motor performance and EF.For example, Roebers and Kauer (2009) controlled for processing speed and found that only a few of the correlations between motor performance and several EF tasks in TD children aged 6-14 years remained significant.Furthermore, processing speed is suggested to be a factor highly important for both motor performance and EF.Global theories of cognitive development conceptualize processing speed as a central mental capacity that drives changes in higher-order cognition, such as EF (Hale, 1990;Kail & Salthouse, 1994).Processing speed and EF are highly intertwined and codependent in preschool children (Clark et al., 2014).In future studies that examine the relationship between motor performance and EF, it will be important to account for the unique and joint contribution of processing speed and to consider it as a possible cause of task impurity.
The high variability in the reported correlations between motor performance and EF may also be partly explained by unknown and uncontrolled child-related and environmental factors.It is not well known how these factors possibly confound and/or moderate the relationship between motor performance and EF.Previous empirical studies have shown that factors, such as age, gender, attention, SES, and ADHD symptomatology attenuated the relationship between motor performance and EF (Houwen et al., 2017;Piek et al., 2008;Wassenberg et al., 2005).In addition, other moderating effects of child-related characteristics have been demonstrated, such as fitness, non-verbal intelligence, reaction time, and visual perception (Aadland et al., 2017;Michel et al., 2018).Another possible explanation is that motor performance and EF may have non-linear (e.g., curvilinear) relationships.However, we are not aware of any studies examining non-linear relationships between both domains.More research is needed to examine more nuanced relationships between motor skills and EF, i.e., paying attention to interactions or non-linear trends.

Strengths and limitations
The major strength of this review lies in its novelty, being the first, to the authors' knowledge, that applied a methodological lens to the study of the relationship between motor performance and EF by outlining the role of EF task response demands.Such a perspective can be important for the development and evaluation of targeted interventions related to motor performance and EF.In addition, this review offers clinicians and researchers with a perspective on the choice of EF measures in the context of clinical and research assessment.
However, some limitations should be considered when interpreting the results.First, a meta-analysis could not be run due to the heterogeneity of EF measures, the limited number of studies in children with MCD, and the lack of power caused by the limited number of correlation coefficients per type of EF task response per motor and EF component.The latter issue also hindered the component-level analysis of the influence of the type of EF task response on the relationship between motor performance and EF.Second, in some of the studies the strength of the nonsignificant correlations was not even reported or only partial correlations, e.g., controlled for age and gender, were stated.Despite our best efforts in requesting these data, some data were not available to be included in the review.Consequently, the impossibility of including these unreported correlation coefficients might have biased the results of this review.Third, although some of the included studies controlled for other variables, only the raw-order correlations between motor performance and EF were included in the analysis to guarantee the most reliable comparison between correlations.However, as a result, valuable information on more pure and of polynomial (e.g., quadratic, cubic, logarithmic) relationships between motor performance and EF and on the role of other variables in their relationship may have been lost.Fourth, only one study (Kim et al., 2018) included an EF task requiring a complex motor response.Therefore, examination of the effect of the level of motor response of EF tasks on the relationship between motor performance and EF was not feasible.Fifth, in the current review, EF was divided into inhibition, working memory, cognitive flexibility, and "other," based on its core components (Diamond, 2013).However, there is an ongoing debate as to whether EF components are already separable in early childhood (Karr et al., 2018).Factor analytic studies on the structure of EF in preschool children have shown inconsistent results.Most studies demonstrated a one-factor structure (Fuhs & Day, 2011;Masten et al., 2012;Shing et al., 2010;Wiebe et al., 2008Wiebe et al., , 2011;;Willoughby et al., 2010Willoughby et al., , 2012)).Other studies identified a two-factor structure with inhibition and working memory or inhibition and working memory / cognitive flexibility as latent components (Lerner & Lonigan, 2014;Miller et al., 2012;Usai et al., 2014) or a three-factor structure with inhibition, working memory, and cognitive flexibility as latent components (Espy et al., 2004;Hughes, 1998;Monette et al., 2011).In empirical studies, EF tasks are often used to examine specific EF components (e.g., Lehmann et al., 2014;Livesey et al., 2006).However, it may be argued whether EF tasks measure single EF components in early childhood.In addition, this may be age-dependent, because EF components seem to progressively separate with age (Nelson et al., 2016).Consequently, the categorization of EF in the current review insinuates that the EF structure at the component-level in the preschool children is challenging to generalize to the youngest preschool children.However, most included studies focused on a relatively large age range.Therefore, it was impossible to take into account different EF factor structures at different ages.We presented the results as comprehensively as possible by presenting the correlations of all motor and EF scores in addition to the correlations on the component-level.For future cross-sectional research on EF, it is recommended to focus on small age ranges.Sixth, the three included studies focusing on children with MCD comprised samples of children without an official diagnosis, i.e., DCD (Alesi et al., 2019;Michel et al., 2011Michel et al., , 2018)).These samples were classified as having motor coordination difficulties when they scored below a certain cutoff score on (a component of) the MABC-2.The European Academy of Childhood Disability does not recommend a diagnosis of DCD before the age of 5 years (Blank et al., 2019), because early motor development is characterized by intra-individual variability (Houwen et al., 2021).Several studies have shown that some preschool children scoring below a certain cutoff score on a standardized assessment of motor performance scored average at a later point in time (Houwen et al., 2021;Michel et al., 2018).Therefore, the current results regarding young children with MCD should be interpreted with caution.Finally, although not a limitation of the review itself, some methodological concerns with regard to the included studies were noted.More than half of the analyzed studies lacked a detailed description of the sample which hindered determining whether a sample should be classified as being typically developing and/or also including developmental problems.In addition, due to a lack of reporting on the inclusion and exclusion criteria, it is often unclear whether the study samples have additional, co-occurring (developmental/medical) problems.Other methodological concerns are that none of the studies justified the sample size and only one study reported the validity of the applied measures.The reliability of the applied measures was mostly not reported.Unfortunately, due to lack of reporting this information in the reviewed articles, we were unable to take into account the potential influence of the use of small samples and lack of reliability and validity of measures on the correlational results.

Future directions
The relationship between motor performance and EF is suggested to be stronger in children with MCD than in children without such difficulties (Rigoli et al., 2013).However, there is a lack of studies examining their direct, correlational relationship in this group.It is recommended to examine the direct relationship between motor performance and EF in children with MCD.
To be able to examine more pure relationships and provide explanations for the highly varying relationships between motor performance and EF that were found, it is recommended to further explore the role of task demands on the relationship between motor performance and EF.Examples of task demands that should be further explored are the complexity of the motor response of EF tasks, the complexity of the actual motor tasks, and additional processes (e.g., memory, intelligence, language comprehension).The role of task demands needs to also be further examined in children with developmental disorders.An example of this is research in more detail on to what extent demands of EF tasks affect the relationship between motor performance and EF in children with DCD, e.g., by manipulating the motor loads within EF tasks to examine whether this affects the relationship between motor performance and EF.This could provide more insight into whether motor coordination difficulties have a general effect on EF or a specific effect on EF only when performing motor tasks.Bearing this knowledge in mind, more optimal EF interventions can be developed for children with motor coordination difficulties, such as DCD.It can, however, be difficult to know how a poor score on a certain task should be interpreted due to two reasons.First, although tasks often aim to tap one specific area, such tasks hardly exist, as also was seen in this review.More specifically, when a child with a developmental disorder scores poorly on a certain task, it is difficult to know whether it is due to poor performance on the process that is intended to be measured or whether the poor score is (also partly) the result of other unintended measured processes (Packwood et al., 2011), such as motor coordination difficulties in children with DCD.Second, symptoms of many developmental disorders (such as ADHD, ASD, and DCD) are not evident in only one area of functioning, but instead, they influence the child's functioning in several domains, i.e., attention, motor performance, and EF (Craig et al., 2016;Flapper & Schoemaker, 2013;Fournier et al., 2010;Kaiser et al., 2015;Pauls & Archibald, 2016;Wilson et al., 2013).Therefore, it is essential for practitioners and researchers to pick tasks carefully by defining precisely what they want to assess and then selecting appropriate assessment tools.When the selected EF tasks also require other additional processes, it is necessary to consider these additional processes required to complete the task.For example, the performance on different tasks measuring the same EF across different domains could be examined (Leonard & Hill, 2014).This could contribute to a better interpretation of a child's EF performance.In addition, further development of motor and EF measures that tap the intended process as purely as possible is vital for furthering our understanding of the underlying mechanisms in developmental disorders.Recently, Sartori et al. (2021) developed a valid inhibition test battery with a variation of responses (i.e., motor or verbal) for 8-to 10-year-old children with DCD.Further research should focus on developing these kinds of test batteries for young children with MCD.Furthermore, as described earlier, since the relationship between motor performance and EF occurs and develops in the context of the individual's bodily interactions with the environment, future research should investigate the influence of confounding and moderating child-related and environmental factors.
A challenge in measuring EF is that executive demands can be age-dependent.The same tasks may therefore tap different skills at different ages (Carlson et al., 2016).EF tasks that are complex at an early age can be simple at a later age.Consequently, EF tasks that are appropriate at a certain age are therefore not necessarily appropriate at another age.As a function of age, it is unfeasible to use the same EF tasks due to floor and ceiling effects (Kurgansky, 2022).For future cross-sectional research on EF, it is recommended to focus on small age ranges or to choose appropriate EF tasks per age.

Conclusion
Our review suggests that highly variable correlations were found for the relationship between motor performance and EF and that the strength of correlations varied mainly from none to moderately positive.Contrary to the expectations, the type of task response of EF tasks did not relate to the strength of relationships between motor performance and EF in 2-to 6-year-old TD children and children with MCD.
It is important to further investigate potential explanations for the highly varying relationships between motor performance and EF, such as the moderating role of task demands in the relationship between motor performance and EF.In the area of motor coordination difficulties, a better understanding of the role of task demands on the relationship between motor performance and EF could lead to the development of fine-tuned differential diagnostic procedures and intervention programs.Note 1.It is not always clear when a child has a motor problem.
The normal acquisition of motor skills shows large variation (Wilson et al., 2013) and it is not always easy to distinguish children who deviate from the normal pattern of motor development.However, this discussion is beyond the current review.
preschools: 47% (F), 53% (M) Rural preschools: 59% (F), 41% (M) Urban preschools: 52.66 mo (SD not reported)Rural preschools: 48.83 mo (SD not reported) coordination disorder; TD: typically developing; F: female; M: male; mo: months; y: years.a reaction time.Stop-signal reaction time is estimated from the average go-signal reaction time taking into account the probability of inhibiting the response at a given Stop-Signal delay., accounting for individual differences in time to task completion under lower inhibitory demands by subtracting time to task completion in the first pages of the task from the high interference correct responses, percentage of trails that successfully responses to target stimulus, percentage of trials that successfully omitting the standard stimulus van der Veer et al. (2020) achtnis Testbatterie; BST: Bivalent Shape Task; BVN5-11: Batteria di Valutazione Neuropsicologica per l'Et a Evolutiva; DCCS: Dimensional Change Card Sort; K-CPT: Conners' Kiddy Continuous Performance Test; HAWIK-IV: Hamburg-Wechsler-Intelligenztest f€ ur Kinder und Jugendliche-4; HTKS: Head-Toes-Knees-Shoulders task; IDS-P: Intelligence and Development Scales-Preschool; KHV-VK: Konzentrations-Handlungsverfahren f€ ur Vorschulkinder; NEPSY: Developmental NEuroPSYchological Assessment; PEBL: Psychology Experiment Building Language; ATT: attention; CF: cognitive flexibility; INH: inhibition; PL: planning; PS: problem-solving; WM: working memory; NA: not applicable.

Figure
Figure2.Correlations between all motor and EF scores per type of EF task response in typically developing children.Note.Positive correlations reflect a positive relationship, that is, better motor performance is related to better EF.Negative correlations reflect a negative relationship, that is, better motor performance is related to worse EF.The absolute values of correlation coefficients above .50or below À.50 were considered as strongly positive or negative, respectively, correlations between .30and .50 and between À.30 and À.50 were considered moderately positive or negative, respectively, correlations between .10 to .30and À.10 to À.30 were considered weakly positive or negative, respectively, correlations equal to or below .10 and equal to or below À.10 were considered as no correlation(Cohen, 1988).Of the EF tasks requiring a motor response, 2.2% showed a weakly negative correlation, 21.0% showed no correlation, 49.7% showed a weakly positive correlation, 23.8% showed a moderately positive correlation, 3.3% showed a strongly positive correlation with motor scores.Of the EF tasks requiring a verbal response, 4.8% showed a weakly negative correlation, 25.8% showed no correlation, 37.1% showed a weakly positive correlation, 24.2% showed a moderately positive correlation, and 8.1% showed a strongly positive correlation with motor scores.

Figure 3 .
Figure 3. Correlations between fine manipulative skills and inhibition and working memory per type of EF task response in typically developing children.Note.Positive correlations reflect a positive relationship, that is, better motor performance is related to better EF.Negative correlations reflect a negative relationship, that is, better motor performance is related to worse EF.The absolute values of correlation coefficients above .50or below À.50 were considered as strongly positive or negative, respectively, correlations between .30and .50 and between À.30 and À.50 were considered moderately positive or negative, respectively, correlations between .10 to .30and À.10 to À.30 were considered weakly positive or negative, respectively, correlations equal to or below .10 and equal to or below À.10 were considered as no correlation(Cohen, 1988).

Figure 4 .
Figure 4. Correlations between gross manipulative skills and inhibition and working memory per type of EF task response in typically developing children.Note.Positive correlations reflect a positive relationship, that is, better motor performance is related to better EF.Negative correlations reflect a negative relationship, that is, better motor performance is related to worse EF.The absolute values of correlation coefficients above .50or below À.50 were considered as strongly positive or negative, respectively, correlations between .30and .50 and between À.30 and À.50 were considered moderately positive or negative, respectively, correlations between .10 to .30and À.10 to À.30 were considered weakly positive or negative, respectively, correlations equal to or below .10 and equal to or below À.10 were considered as no correlation(Cohen, 1988).

Figure 5 .
Figure5.Correlations between stability and inhibition and working memory per type of EF task response in typically developing children.Note.Positive correlations reflect a positive relationship, that is, better motor performance is related to better EF.Negative correlations reflect a negative relationship, that is, better motor performance is related to worse EF.The absolute values of correlation coefficients above .50or below À.50 were considered as strongly positive or negative, respectively, correlations between .30and .50 and between À.30 and À.50 were considered moderately positive or negative, respectively, correlations between .10 to .30and À.10 to À.30 were considered weakly positive or negative, respectively, correlations equal to or below .10 and equal to or below À.10 were considered as no correlation(Cohen, 1988).

Table 3 .
Characteristics of the motor measures.